With Question/Answer Animations. Chapter 1, Part III: Proofs

Transcription

1 With Question/Answer Animations Chapter 1, Part III: Proofs

2 Summary Valid Arguments and Rules of Inference Proof Methods Proof Strategies

3 Section 1.6

4 Section Summary Valid Arguments Inference Rules for Propositional Logic Using Rules of Inference to Build Arguments Rules of Inference for Quantified Statements Building Arguments for Quantified Statements

5 Revisiting the Socrates Example We have the two premises: All men are mortal. Socrates is a man. And the conclusion: Socrates is mortal. How do we get the conclusion from the premises?

6 The Argument We can express the premises (above the line) and the conclusion (below the line) in predicate logic as an argument: We will see shortly that this is a valid argument.

7 Valid Arguments We will show how to construct valid arguments in two stages; first for propositional logic and then for predicate logic. The rules of inference are the essential building block in the construction of valid arguments. 1. Propositional Logic Inference Rules 2. Predicate Logic Inference rules for propositional logic plus additional inference rules to handle variables and quantifiers.

8 Arguments in Propositional Logic A argument in propositional logic is a sequence of propositions. All but the final proposition are called premises. The last statement is the conclusion. The argument is valid if the premises imply the conclusion. An argument form is an argument that is valid no matter what propositions are substituted into its propositional variables. If the premises are p 1,p 2,,p n and the conclusion is q then (p 1 p 2 p n ) q is a tautology. Inference rules are all argument simple argument forms that will be used to construct more complex argument forms.

9 Rules of Inference for Propositional Logic: Modus Ponens Corresponding Tautology: (p (p q)) q Example: Let p be It is snowing. Let q be I will study discrete math. If it is snowing, then I will study discrete math. It is snowing. Therefore, I will study discrete math.

10 Modus Tollens Corresponding Tautology: ( p (p q)) q Example: Let p be it is snowing. Let q be I will study discrete math. If it is snowing, then I will study discrete math. I will not study discrete math. Therefore, it is not snowing.

11 Hypothetical Syllogism Corresponding Tautology: ((p q) (q r)) (p r) Example: Let p be it snows. Let q be I will study discrete math. Let r be I will get an A. If it snows, then I will study discrete math. If I study discrete math, I will get an A. Therefore, If it snows, I will get an A.

12 Disjunctive Syllogism Corresponding Tautology: ( p (p q)) q Example: Let p be I will study discrete math. Let q be I will study English literature. I will study discrete math or I will study English literature. I will not study discrete math. Therefore, I will study English literature.

13 Addition Corresponding Tautology: p (p q) Example: Let p be I will study discrete math. Let q be I will visit Las Vegas. I will study discrete math. Therefore, I will study discrete math or I will visit Las Vegas.

14 Simplification Corresponding Tautology: (p q) p Example: Let p be I will study discrete math. Let q be I will study English literature. I will study discrete math and English literature Therefore, I will study discrete math.

15 Conjunction Corresponding Tautology: ((p) (q)) (p q) Example: Let p be I will study discrete math. Let q be I will study English literature. I will study discrete math. I will study English literature. Therefore, I will study discrete math and I will study English literature.

16 Resolution Resolution plays an important role in AI and is used in Prolog. Corresponding Tautology: (( p r) (p q)) (q r) Example: Let p be I will study discrete math. Let r be I will study English literature. Let q be I will study databases. I will not study discrete math or I will study English literature. I will study discrete math or I will study databases. Therefore, I will study databases or I will English literature.

17 Using the Rules of Inference to Build Valid Arguments A valid argument is a sequence of statements. Each statement is either a premise or follows from previous statements by rules of inference. The last statement is called conclusion. A valid argument takes the following form: S 1 S 2... S n C

18 Valid Arguments Example 1: From the single proposition Show that q is a conclusion. Solution:

19 Valid Arguments Example 2: With these hypotheses: It is not sunny this afternoon and it is colder than yesterday. We will go swimming only if it is sunny. If we do not go swimming, then we will take a canoe trip. If we take a canoe trip, then we will be home by sunset. Using the inference rules, construct a valid argument for the conclusion: We will be home by sunset. Solution: 1. Choose propositional variables: p : It is sunny this afternoon. r : We will go swimming. t : We will be home by sunset. q : It is colder than yesterday. s : We will take a canoe trip. 2. Translation into propositional logic: Continued on next slide

20 Valid Arguments 3. Construct the Valid Argument

21 Handling Quantified Statements Valid arguments for quantified statements are a sequence of statements. Each statement is either a premise or follows from previous statements by rules of inference which include: Rules of Inference for Propositional Logic Rules of Inference for Quantified Statements The rules of inference for quantified statements are introduced in the next several slides.

22 Universal Instantiation (UI) Example: Our domain consists of all dogs and Fido is a dog. All dogs are cuddly. Therefore, Fido is cuddly.

23 Universal Generalization (UG) Used often implicitly in Mathematical Proofs.

24 Existential Instantiation (EI) Example: There is someone who got an A in the course. Let s call her aand say that agot an A

25 Existential Generalization (EG) Example: Michelle got an A in the class. Therefore, someone got an A in the class.

26 Using Rules of Inference Example 1: Using the rules of inference, construct a valid argument to show that John Smith has two legs is a consequence of the premises: Every man has two legs. John Smith is a man. Solution: Let M(x) denote x is a man and L(x) x has two legs and let John Smith be a member of the domain. Valid Argument:

27 Using Rules of Inference Example 2: Use the rules of inference to construct a valid argument showing that the conclusion Someone who passed the first exam has not read the book. follows from the premises A student in this class has not read the book. Everyone in this class passed the first exam. Solution: Let C(x) denote xis in this class, B(x) denote x has read the book, and P(x) denote xpassed the first exam. First we translate the premises and conclusion into symbolic form. Continued on next slide

28 Using Rules of Inference Valid Argument:

29 Returning to the Socrates Example

30 Solution for Socrates Example Valid Argument

31 Universal Modus Ponens Universal Modus Ponens combines universal instantiation and modus ponens into one rule. This rule could be used in the Socrates example.

32 Section 1.7

33 Section Summary Mathematical Proofs Forms of Theorems Direct Proofs Indirect Proofs Proof of the Contrapositive Proof by Contradiction

34 Proofs of Mathematical Statements A proof is a valid argument that establishes the truth of a statement. In math, CS, and other disciplines, informal proofs which are generally shorter, are generally used. More than one rule of inference are often used in a step. Steps may be skipped. The rules of inference used are not explicitly stated. Easier for to understand and to explain to people. But it is also easier to introduce errors. Proofs have many practical applications: verification that computer programs are correct establishing that operating systems are secure enabling programs to make inferences in artificial intelligence showing that system specifications are consistent

35 Definitions A theorem is a statement that can be shown to be true using: definitions other theorems axioms (statements which are given as true) rules of inference A lemma is a helping theorem or a result which is needed to prove a theorem. A corollary is a result which follows directly from a theorem. Less important theorems are sometimes called propositions. A conjectureis a statement that is being proposed to be true. Once a proof of a conjecture is found, it becomes a theorem. It may turn out to be false.

36 Forms of Theorems Many theorems assert that a property holds for all elements in a domain, such as the integers, the real numbers, or some of the discrete structures that we will study in this class. Often the universal quantifier (needed for a precise statement of a theorem) is omitted by standard mathematical convention. For example, the statement: If x > y, where xand yare positive real numbers, then x 2 > y 2 really means For all positive real numbers xand y, if x > y, then x 2 > y 2.

37 Proving Theorems Many theorems have the form: To prove them, we show that where c is an arbitrary element of the domain, By universal generalization the truth of the original formula follows. So, we must prove something of the form:

38 Proving Conditional Statements: p Trivial Proof: If we know q is true, then p q is true as well. If it is raining then 1=1. Vacuous Proof: If we know p is false then p q is true as well. If I am both rich and poor then = 5. [ Even though these examples seem silly, both trivial and vacuous proofs are often used in mathematical induction, as we will see in Chapter 5) ]

39 Even and Odd Integers Definition: The integer n is even if there exists an integer ksuch that n = 2k, and n is odd if there exists an integer k, such that n = 2k + 1. Note that every integer is either even or odd and no integer is both even and odd. We will need this basic fact about the integers in some of the example proofs to follow. We will learn more about the integers in Chapter 4.

40 Proving Conditional Statements: p Direct Proof: Assume that p is true. Use rules of inference, axioms, and logical equivalences to show that q must also be true. Example: Give a direct proof of the theorem If n is an odd integer, then n 2 is odd. Solution: Assume that n is odd. Then n = 2k + 1for an integer k. Squaring both sides of the equation, we get: n 2 = (2k + 1) 2 = 4k 2 + 4k+1 = 2(2k 2 + 2k) + 1= 2r+ 1, where r= 2k 2 + 2k, an integer. We have proved that if nis an odd integer, then n 2 is an odd integer. ( marks the end of the proof. Sometimes QED is used instead. )

41 Proving Conditional Statements: p Definition: The real number r is rational if there exist integers pand qwhere q 0 such that r = p/q Example: Prove that the sum of two rational numbers is rational. Solution: Assume r and sare two rational numbers. Then there must be integers p, q and also t, u such that Thus the sum is rational. where v = pu+qt w = qu 0

42 Proving Conditional Statements: p Proof by Contraposition: Assume q and show p is true also. This is sometimes called an indirect proof method. If we give a direct proof of q p then we have a proof of p q. Why does this work? Example: Prove that if n is an integer and 3n + 2is odd, thenn is odd. Solution: Assume n is even. So, n = 2k for some integer k. Thus 3n + 2 = 3(2k) + 2 =6k+2 = 2(3k+ 1) = 2j for j = 3k +1 Therefore 3n + 2 is even. Since we have shown q p, p q must hold as well. If n is an integer and 3n + 2is odd (not even), thenn is odd (not even).

43 Proving Conditional Statements: p Example: Prove that for an integer n, if n 2 is odd, then n is odd. Solution: Use proof by contraposition. Assume n is even (i.e., not odd). Therefore, there exists an integer ksuch that n = 2k. Hence, n 2 = 4k 2 = 2 (2k 2 ) and n 2 is even(i.e., not odd). We have shown that if n is an even integer, then n 2 is even. Therefore by contraposition, for an integern, if n 2 is odd, then n is odd.

44 Proving Conditional Statements: p Proof by Contradiction: (AKA reductio ad absurdum). To prove p, assume p and derive a contradiction such as p p. (an indirect form of proof). Since we have shown that p Fis true, it follows that the contrapositive T p also holds. Example: Prove that if you pick 22 days from the calendar, at least 4 must fall on the same day of the week. Solution: Assume that no more than 3 of the 22 days fall on the same day of the week. Because there are 7 days of the week, we could only have picked 21 days. This contradicts the assumption that we have picked 22 days.

45 Proof by Contradiction A preview of Chapter 4. Example: Use a proof by contradiction to give a proof that 2 is irrational. Solution: Suppose 2 is rational. Then there exists integers a and b with 2 = a/b, where b 0 and aand b have no common factors (see Chapter 4). Then Therefore a 2 must be even. If a 2 is even then amust be even (an exercise). Since ais even, a = 2c for some integer c. Thus, Therefore b 2 is even. Again then bmust be even as well. But then 2 must divide both aand b. This contradicts our assumption that aand bhave no common factors. We have proved by contradiction that our initial assumption must be false and therefore 2 is irrational.

46 Proof by Contradiction A preview of Chapter 4. Example: Prove that there is no largest prime number. Solution: Assume that there is a largest prime number. Call it p n. Hence, we can list all the primes 2,3,.., p n. Form None of the prime numbers on the list divides r. Therefore, by a theorem in Chapter 4, either r is prime or there is a smaller prime that divides r. This contradicts the assumption that there is a largest prime. Therefore, there is no largest prime.

47 Theorems that are Biconditional Statements To prove a theorem that is a biconditionalstatement, that is, a statement of the form p q, we show that p qand q pare both true. Example: Prove the theorem: If nis an integer, then nis odd if and only if n 2 is odd. Solution: We have already shown (previous slides) that both p qand q p. Therefore we can conclude p q. Sometimes iff is used as an abbreviation for if an only if, as in If nis an integer, then nis odd iifn 2 is odd.

48 What is wrong with this? Proof that 1 = 2 Solution: Step 5. a - b = 0 by the premise and division by 0 is undefined.

49 Looking Ahead If direct methods of proof do not work: We may need a clever use of a proof by contraposition. Or a proof by contradiction. In the next section, we will see strategies that can be used when straightforward approaches do not work. In Chapter 5, we will see mathematical induction and related techniques. In Chapter 6, we will see combinatorial proofs

50 Section 1.8

51 Section Summary Proof by Cases Existence Proofs Constructive Nonconstructive Disproof by Counterexample Nonexistence Proofs Uniqueness Proofs Proof Strategies Proving Universally Quantified Assertions Open Problems

52 Proof by Cases To prove a conditional statement of the form: Use the tautology Each of the implications is a case.

53 Proof by Cases Example: Let b = max{a, b} = a if a b, otherwise b = max{a, b} = b. Show that for all real numbers a, b, c = c) (This means the is associative.) Proof: Let a, b, and c be arbitrary real numbers. Then one of the following 6 cases must hold. 1. a b c 2. a c b 3. b a c 4. b c a 5. c a b 6. c b a Continued on next slide

54 Proof by Cases Case 1: a b c (a@ b) = a, c = a, c = b Hence (a@ c = a = (b@ c) Therefore the equality holds for the first case. A complete proof requires that the equality be shown to hold for all 6 cases. But the proofs of the remaining cases are similar. Try them.

55 Without Loss of Generality Example: Show that if xand yare integers and both x y and x+yare even, then both xand yare even. Proof: Use a proof by contraposition. Suppose x and yare not both even. Then, one or both are odd. Without loss of generality, assume that x is odd. Then x= 2m+ 1 for some integer k. Case 1: y is even. Then y= 2nfor some integer n, so x+y= (2m+ 1) + 2n= 2(m+ n) + 1 is odd. Case 2:yis odd. Then y= 2n +1 for some integer n, so x y= (2m+ 1) (2n+ 1) = 2(2m n+m+ n) + 1 is odd. We only cover the case where xis odd because the case where yis odd is similar. The use phrase without loss of generality(wlog) indicates this.

56 Existence Proofs Proof of theorems of the form. Constructiveexistence proof: Find an explicit value of c, for which P(c) is true. Then is true by Existential Generalization (EG). Example: Show that there is a positive integer that can be written as the sum of cubes of positive integers in two different ways: Proof: 1729 is such a number since 1729 = = Godfrey Harold Hardy ( ) Srinivasa Ramanujan ( )

57 NonconstructiveExistence Proofs In a nonconstructiveexistence proof, we assume no c exists which makes P(c)true and derive a contradiction. Example: Show that there exist irrational numbers x and ysuch that x y is rational. Proof:We know that 2 is irrational. Consider the number 2 2. If it is rational, we have two irrational numbers x and y with x y rational, namely x = 2 and y= 2. But if 2 2 is irrational, then we can let x= 2 2 and y= 2 so that aaaaa x y =( 2 2 ) 2 = 2 ( 2 2) = 2 2 = 2.

58 Counterexamples Recall. To establish that is true (or is false) find a csuch that P(c) is true or P(c) is false. In this case c is called a counterexampleto the assertion. Example: Every positive integer is the sum of the squares of 3 integers. The integer 7 is a counterexample. So the claim is false.

59 Uniqueness Proofs Some theorems asset the existence of a unique element with a particular property,!x P(x). The two parts of a uniqueness proof are Existence: We show that an element xwith the property exists. Uniqueness: We show that if y x, then ydoes not have the property. Example: Show that if aand bare real numbers and a 0, then there is a unique real number r such that ar+ b= 0. Solution: Existence: The real number r= b/ais a solution of ar+ b= 0 because a( b/a) + b= b+ b=0. Uniqueness: Suppose that sis a real number such that as + b= 0. Then ar+ b= as+ b, where r= b/a. Subtracting b from both sides and dividing by ashows that r= s.

60 Proof Strategies for proving p Choose a method. 1. First try a direct method of proof. 2. If this does not work, try an indirect method (e.g., try to prove the contrapositive). For whichever method you are trying, choose a strategy. 1. First try forward reasoning. Start with the axioms and known theorems and construct a sequence of steps that end in the conclusion. Start with pand prove q, or start with q and prove p. 2. If this doesn t work, try backward reasoning. When trying to prove q, find a statement p that we can prove with the property p q.

61 Backward Reasoning Example: Suppose that two people play a game taking turns removing, 1, 2, or 3 stones at a time from a pile that begins with 15stones. The person who removes the last stone wins the game. Show that the first player can win the game no matter what the second player does. Proof: Let n be the last step of the game. Step n: Player 1 can win if the pile contains 1,2, or 3stones. Step n-1: Player 2 will have to leave such a pile if the pile that he/she is faced with has 4 stones. Step n-2: Player 1 can leave 4 stones when there are 5,6, or 7stones left at the beginning of his/her turn. Step n-3: Player 2 must leave such a pile, if there are 8 stones. Step n-4: Player 1 has to have a pile with 9,10, or 11stones to ensure that there are 8 left. Step n-5: Player 2 needs to be faced with 12stones to be forced to leave 9,10,or 11. Step n-6: Player 1 can leave 12stones by removing 3stones. Now reasoning forward, the first player can ensure a win by removing 3stones and leaving 12.

62 Universally Quantified Assertions To prove theorems of the form,assume x is an arbitrary member of the domain and show that P(x) must be true. Using UG it follows that. Example: An integerx is even if and only if x 2 is even. Solution: The quantified assertion is x [x is even x 2 is even] We assume x is arbitrary. Recall that is equivalent to So, we have two cases to consider. These are considered in turn. Continued on next slide

63 Universally Quantified Assertions Case 1. We show that if x is even then x 2 is even using a direct proof (the only if part or necessity). If x is even then x = 2k for some integer k. Hence x 2 = 4k 2 = 2(2k 2 ) which is even since it is an integer divisible by 2. This completes the proof of case 1. Case 2on next slide

64 Universally Quantified Assertions Case 2. We show that if x 2 is even then x must be even (the if part or sufficiency). We use a proof by contraposition. Assume x is not even and then show that x 2 is not even. If x is not even then it must be odd. So, x= 2k+ 1 for some k. Then x 2 = (2k + 1) 2 = 4k 2 + 4k+ 1 = 2(2k 2 + 2k) + 1 which is odd and hence not even. This completes the proof of case 2. Since xwas arbitrary, the result follows by UG. Therefore we have shown that x is even if and only if x 2 is even.

65 Proof and Disproof: Tilings Example 1: Can we tile the standard checkerboard using dominos? Solution: Yes! One example provides a constructive existence proof. Two Dominoes The Standard Checkerboard One Possible Solution

66 Tilings Example 2: Can we tile a checkerboard obtained by removing one of the four corner squares of a standard checkerboard? Solution: Our checkerboard has 64 1 = 63squares. Since each domino has two squares, a board with a tiling must have an even number of squares. The number 63 is not even. We have a contradiction.

67 Tilings Example 3: Can we tile a board obtained by removing both the upper left and the lower right squares of a standard checkerboard? Nonstandard Checkerboard Dominoes Continued on next slide

68 Tilings Solution: There are 62 squares in this board. To tile it we need 31 dominos. Key fact: Each domino covers one black and one white square. Therefore the tiling covers 31 black squares and 31 white squares. Our board has either 30 black squares and 32 white squares or 32 black squares and 30 white squares. Contradiction!

69 The Role of Open Problems Unsolved problems have motivated much work in mathematics. Fermat s Last Theorem was conjectured more than 300 years ago. It has only recently been finally solved. Fermat s Last Theorem: The equation x n + y n = z n has no solutions in integers x, y, and z, with xyz 0 whenever n is an integer with n> 2. A proof was found by Andrew Wiles in the 1990s.

70 An Open Problem The 3x + 1 Conjecture: Let T be the transformation that sends an even integer xto x/2 and an odd integer xto 3x + 1. For all positive integers x, when we repeatedly apply the transformation T, we will eventually reach the integer 1. For example, starting with x = 13: T(13) = = 40, T(40) = 40/2 = 20, T(20) = 20/2 = 10, T(10) = 10/2 = 5, T(5) = = 16,T(16) = 16/2 = 8, T(8) = 8/2 = 4, T(4) = 4/2 = 2, T(2) = 2/2 = 1 The conjecture has been verified using computers up to

71 Additional Proof Methods Later we will see many other proof methods: Mathematical induction, which is a useful method for proving statements of the form n P(n), where the domain consists of all positive integers. Structural induction, which can be used to prove such results about recursively defined sets. Cantor diagonalizationis used to prove results about the size of infinite sets. Combinatorial proofs use counting arguments.